Test report DSG-4670 · Rev A · tested November 7, 2026
Supply Chain & PolicyDevice under test
Intel’s $6.1bn substrate deal exposes AI packaging crunch
Intel committed $5.3bn-$6.1bn to substrates, while Samsung plans ₩6.78tn of FC-BGA capacity and LG Innotek targets 2027 AI production as Intel expects substrate shortages to persist through next year.
- Read
- 3 min
- Words
- 678
- Node
- 5nm
- Operator
- Amara Osei
Spec summary
- Intel finalized two substrate contracts worth $5.3 billion to $6.1 billion; $3.3 billion is due by end-2027.
- Intel expects substrate and memory shortages to persist through next year.
- Samsung’s ₩6.78 trillion FC-BGA expansion pairs a September 2028 Sejong start with June 2030 in Vietnam.
- Samsung Electro-Mechanics’ third-quarter package solution revenue rose 58% to ₩939.0 billion.
- LG Innotek is discussing a 100mm-plus AI FC-BGA, with mass production targeted for 2027 and server-grade output by 2028.

By Amara Osei
Intel has finalized two semiconductor substrate supply contracts with expected outlays of $5.3 billion to $6.1 billion, with about $3.3 billion scheduled for disbursement by the end of 2027. In won terms, the commitments equal roughly ₩7.1 trillion to ₩8.2 trillion.
That spending shows a bottleneck extending beyond wafer fabrication. Intel disclosed in its third-quarter filing with the US Securities and Exchange Commission that it cannot meet sufficient demand for data-center central processing units because of a shortage of packaging substrates. Industry sources on the 4th reported the disclosure.
How large is Intel’s substrate commitment?
Intel expects the supply crunch for substrates and memory to persist through next year. The contracts form part of a wider race among global chipmakers to secure advanced packaging components, and their payment schedules show manufacturers committing capital years before using all of it.
Takeaway: Intel is committing billions now to secure components it will use in later years.
Why does higher AI chip performance strain substrate supply?
AI servers use CPUs and GPUs with higher performance, and the substrates connecting them must grow larger and more complex. Manufacturers must stack multiple circuit layers with precision, which raises production difficulty. Adding factory capacity takes significant time, so chip performance gains translate into longer lead times for packaging components.
The flip-chip ball grid array, or FC-BGA, is a high-value-added substrate that transmits electrical signals between chips and mainboards. AI servers and data centers use the format widely. As the shortage tightens, some customers are subsidizing substrate manufacturers’ facility investment costs and guaranteeing long-term purchase volumes.
Takeaway: More complex AI processors require more substrate area and precision, so capacity cannot expand as quickly as demand.
What is Samsung Electro-Mechanics adding?
Samsung Electro-Mechanics announced ₩6.78 trillion of investment in September to expand FC-BGA production at its Sejong campus and Vietnam subsidiary. It assigned ₩4.27 trillion to Sejong and ₩2.51 trillion to Vietnam. The Sejong facility is scheduled for mass production in September 2028, while the Vietnam plant is planned to start in June 2030.
Global clients provide financial backing and mid-to-long-term purchase commitments for the expansion. That arrangement reduces Samsung Electro-Mechanics’ investment burden while securing future sales volumes.
Customer orders continued after the September announcement. On its third-quarter earnings conference call, Samsung Electro-Mechanics said: "The currently planned expansions alone would be insufficient to meet all customer demand." The company is reviewing the timing and scale of additional investment.
Third-quarter package solution revenue reached ₩939.0 billion, up 58% from a year earlier. Sales of high-value FC-BGA for server CPUs, AI accelerators and network equipment drove the increase.
Takeaway: Samsung is expanding FC-BGA output, but current projects already fall short of customer demand.
What is LG Innotek bringing to the AI substrate market?
LG Innotek unveiled an AI accelerator-focused FC-BGA with a side length exceeding 100 millimeters last September. The company is discussing supply plans with customers and aims to begin mass production for AI training and inference in 2027. By 2028, it plans to extend into server-grade, high-value substrates.
LG Innotek is negotiating long-term supply agreements and additional investment schemes with two major clients. Industry watchers describe the competition as extending beyond chip manufacturing to securing stable supplies of core components.
Takeaway: LG Innotek’s customer negotiations remain underway ahead of the 2027 production target.
How will the shortage affect substrate makers?
Kwon Min-gyu, a researcher at Kiwoom Securities, said: "As AI semiconductor demand rapidly grows, substrate supply is failing to keep pace." He expects ongoing supply shortages to support earnings for substrate manufacturers for the time being.
For chipmakers, the immediate task is turning capital into substrate capacity on multi-year schedules. For substrate makers, customer financing and long-term purchase commitments reduce investment burdens and pre-secure sales volumes.
Takeaway: For now, substrate makers stand to benefit from shortages, while chipmakers wait years for added capacity.
via Google News: Semiconductor supply chain (Source)
More from Amara Osei
Show full bio
Staff writer covering money, funding and market shifts around hardware layer of artificial intelligence at Die Signal.
593 articles
Same lot · LOT-360E
- DSG-805645nmIntel’s $5.3–$6.1 Billion Substrate Plan Faces AI Shortage
- DSG-952510nmChina's Xinai raises $224M to scale AI chip substrate output
- DSG-30177nmApplied Materials Puts $5 Billion Into Silicon Valley AI Chip Research Hub
- DSG-301820nmTSMC Holds the Cards in AI Chip Capacity Fight
- DSG-262410nmApplied Materials, Intel Widen AI Chip Development Partnership